定制VLIW结构实现四精度浮点基本函数

doi:10.3969/j.issn.0372-2112.2012.09.003

电子学报 ›› 2012, Vol. 40 ›› Issue (9): 1715-1722.DOI: 10.3969/j.issn.0372-2112.2012.09.003

定制VLIW结构实现四精度浮点基本函数

雷元武, 窦勇, 倪时策, 周杰

国防科学技术大学计算机学院, 湖南长沙 410073

收稿日期:2011-03-18 修回日期:2011-05-03 出版日期:2012-09-25
- 作者简介:
- 雷元武 男,1982年1月出生于湖南桂阳,2007年于国防科技大学获计算机科学与技术专业硕士学位,现为08级博士研究生.主要研究方向:高性能计算机系统结构. E-mail:yuanwulei@nudt.edu.cn
- 基金资助:
- 国家自然科学基金 (No.61125201)

A Special-Purpose VLIW Structure to Implement Quadruple Precision Floating-Point Elementary Functions

LEI Yuan-wu, DOU Yong, NI Shi-ce, ZHOU Jie

School of Computer, National University of Defense Technology, Changsha, Hunan 410073, China

Received:2011-03-18 Revised:2011-05-03 Online:2012-09-25 Published:2012-09-25
- Supported by:
- National Natural Science Foundation of China (No.61125201)

摘要/Abstract

摘要： 本文针对科学应用中基本函数种类多、实现复杂、使用频率低的特点,提出一种定制VLIW结构四精度浮点基本函数协处理器(QPC-Processor).该结构通过显示并行技术挖掘基本函数实现算法的并行性,在同一硬件平台上通过元操作的不同组合来计算多种基本函数.同时,本文还提出基本函数元操作序列到定制VLIW指令的映射算法,指导基本函数的设计.最后,在FPGA平台上进行验证.实验结果表明,相对软件实现,单个QPC-Processor能够取得6倍以上的加速比,而且,QFC-Processor在同一硬件平台上实现多种类型的算法,弥补单一算法的不足,获得较高的硬件资源利用率.

关键词: 四精度浮点算术, 超长指令字, 基本函数, CORDIC算法

Abstract: This paper proposes a quadruple precision floating-point elementary function co-processor based on very large instruction word (VLIW) structure (QPC-Processor),which exploits the parallelism through the explicitly parallel technology of the VLIW structure.Variety of quadruple precision elementary functions is evaluated via the different combination of basic operation in the unified hardware.Finally,we prototype the QPC-Processor units into FPGA chip.The experimental results show our design outperforms the software approach by a factor of more than 6.Moreover,high utilization of hardware resource can be obtained in QFC-Processor.

Key words: quadruple precision floating-point arithmetic, very long instruction word (VLIW), elementary function, CORDIC algorithm

中图分类号:

TN302

雷元武, 窦勇, 倪时策, 等. 定制VLIW结构实现四精度浮点基本函数[J]. 电子学报, 2012, 40(9): 1715-1722.

LEI Yuan-wu, DOU Yong, NI Shi-ce, et al. A Special-Purpose VLIW Structure to Implement Quadruple Precision Floating-Point Elementary Functions[J]. Acta Electronica Sinica, 2012, 40(9): 1715-1722.

参考文献

[1] Y Dou,Y Lei,G Wu,et al.FPGA accelerating double/quad-double high precision floating-point application for exascale computing. Proceedings of 24th International Conference on Supercomputing. Tsukuba:ACM Press,2010.325-336.
[2] D H Bailey.High-precision floating-point arithmetic in scientific computation[J].Computing in Science and Engineering,2005,7(3):54-61.
[3] ANSI/IEEE Std 754-2008, Standard for Binary Floating Point Arithmetic ANSI/IEEE Standard 754-2008[S].
[4] L Fousse,G Hanrot,V Lefevre,P Pelissier.MPFR:A multiple-precision binary floating-point library with correct rounding[J].Transactions on Mathematical Software,2007,33(2):1-15.
[5] Y Hida,X S Li,D H Bailey.Quad-Double Arithmetic:Algorithms,Implementation,and Application. Berkeley:Lawrence Berkeley National Laboratory,2000.
[6] E M Schwarz,R M Smth,C A Krygowski.The S/390 G5 floating point unit supporting hex and binary architectures. Proceedings of the 14th IEEE Symposium on Computer Arithmetic. Adelaide:IEEE Press,1999.836-841.
[7] A Akkas.Dual-mode quadruple precision floating-point adder. Proceedings of 9th Euromicro Conference on Digital System Design. Dubrovnik:IEEE Press,2006.211-220.
[8] A Akkas,M Schult.Dual-mode floating-point multiplier architectures with parallel operations[J].Journal of Systems Architecture,2006,52(10):549-562.
[9] A Isseven,A Akkas.A dual-mode quadruple precision floating-point divider. Proceedings of Fortieth Asilomar Conference on Signals,Systems and Computers. Pacific Grove:IEEE Press,2006.1697-1701.
[10] Ej-Araby,E Gonzalez,I El-Ghazawi.Bringing high performance reconfigurable computing to exact computations. Proceedings of International Conference on Field Programmable Logic and Applications. Amsterdam:IEEE Press,2007.79-85.
[11] J Detrey,F Dinechin,X Pujol.Return of the hardware floating-point elementary function. Proceedings of the 18th IEEE Symposium on Computer Arithmetic. Montepellier:IEEE Press,2007.161-168.
[12] K K Parhi,H R Srinivas.A fast radix-4 division algorithm and its architecture. IEEE Transactions on Computers,1995,44(6):826-831.
[13] Y Li,W Chu.Parallel-array implementations of a non-restoring square root algorithm. IEEE International Conference on Computer Design. Austin:IEEE Press,1997.690-695.
[14] J S Walther.A unified algorithm for elementary functions. Proceedings of AFIPS Joint Computer Conference. New York:IEEE Press,1971.379-385.
[15] P Soderquist,M Leeser.Division and square root:Choosing the right implementation[J].IEEE Micro,1997,17(4):56-66.
[16] M D Ercegovac,L Imbert,et al.Improving goldschmidt division,square root and square root reciprocal[J].IEEE Transactions on Computers,2000,49(7):759-763.
[17] J Zhou,Y Dou,Y Lei,et al.Double precision Hybrid mode floating-point FPGA CORDIC co-processor. Proceedings of the 10th IEEE International Conference on High Computing and Communications. Dalin:IEEE Press,2008.182-189.

定制VLIW结构实现四精度浮点基本函数

A Special-Purpose VLIW Structure to Implement Quadruple Precision Floating-Point Elementary Functions

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 4

编辑推荐

Metrics

[1]	徐成, 秦云川, 李肯立, 戚芳芳. 免缩放因子双步旋转CORDIC算法[J]. 电子学报, 2014, 42(7): 1441-1445.
[2]	沈钲;孙义和. 一种支持同时多线程的VLIW DSP架构[J]. 电子学报, 2010, 38(2): 352-358.
[3]	欧阳万里, 肖创柏, 刘广. 一种新的基于VLIW的IDCT和运动补偿算法[J]. 电子学报, 2005, 33(11): 2074-2079.
[4]	李学明;李继. 用超长指令实现DCT的新算法[J]. 电子学报, 2003, 31(7): 1074-1077.